We demonstrate with a nonlinear MHD code that angular momentum can be transported due to the magnetic instability of toroidal fields under the influence of differential rotation, and that the resulting effective viscosity may be high enough to explain the almost rigid-body rotation observed in radiative stellar cores. Only stationary current-free fields and only those combinations of rotation rates and magnetic field amplitudes which provide maximal numerical values of the viscosity are considered. We find that the dimensionless ratio of the effective over molecular viscosity, $\nu_T/\nu$;, linearly grows with the Reynolds number of the rotating fluid multiplied with the square-root of the magnetic Prandtl number - which is of order unity for the considered red sub-giant KIC 7341231. For the considered interval of magnetic Reynolds numbers - which is restricted by numerical constraints of the nonlinear MHD code - there is a remarkable influence of the magnetic Prandtl number on the relative importance of the contributions of the Reynolds stress and the Maxwell stress to the total viscosity, which is magnetically dominated only for Pm $\gtrsim$ 0.5. We also find that the magnetized plasma behaves as a non-Newtonian fluid, i.e. the resulting effective viscosity depends on the shear in the rotation law. The decay time of the differential rotation thus depends on its shear and becomes longer and longer during the spin-down of a stellar core.

The empirical finding with data from the Solar Dynamics Observatory (SDO) of positive (negative) horizontal Reynolds stress at the northern (southern) hemisphere for solar giant cells (Hathaway et al. 2013) is discussed for its consequences for the theory of the solar/stellar differential rotation. Solving the nonlinear Reynolds equation for the angular velocity under neglect of the meridional circulation we show that the horizontal Reynolds stress of the northern hemisphere is always negative at the surface but it is positive in the bulk of the solar convection zone by the action of the Lambda effect. The Lambda effect, which describes the angular momentum transport of rigidly rotating anisotropic turbulence and which avoids a rigid-body rotation of the convection zones, is in horizontal direction of cubic power in the rotation rate $\Omega$ and it is always equatorwards directed. Theories without Lambda effect which may also provide the observed solar rotation law only by the action of a meridional circulation lead to a horizontal Reynolds stress with the opposite sign as observed.